Bea Wolf (review), by Zach Weinersmith (author) and Boulet (illustrator)

18 April 2023

Weinersmith, Zach (author) and Boulet (illustrator). Bea Wolf. New York: First Second, 2023. 208 pages. US$19.99.

Listen to the lives of the long-ago kids, the world-fighters
The parent-unminding kids, the improper,
The politeness-proof,
The unbowed bully-crushers,
The bedtime-breakers, the raspberry-blowers,
Fighters of fun-killers, fearing nothing, fated for fame.

There was Tanya, treat-taker, terror of Halloween,
Her costume-cache vast, sieging kin and neighbor,
Draining full candy-bins, fearing not the fate of her teeth.
Ten thousand treats she took. That was a fine Tuesday.

So begins the mock-epic tale of Bea Wolf, by Zach Weinersmith and Boulet, a delightful adaptation of the Old English epic Beowulf. This latest adaptation of the classic tale will delight children and adults alike.

In it, the titular hero has been transformed into a five-year-old girl and placed into a world of childhood imaginative exaggeration. A world filled with treasure troves of Halloween candy, brave battles with the kids from the next block over, and a fearsome adult-monster, Grindle, who drives King Karl and his gang from their treehouse, Treeheart, with his power to turn children into teenagers. And striding into the scene to rescue Karl and his prepubescent courtiers is Bea Wolf from Heidi’s Hold across the river.

Weinersmith has not only captured the tone and rhythm of the original epic, filling it with alliteration and newly coined kennings for the twenty-first century, but he has also captured the elegiac mood of the Old English poem, for lurking in the background is the knowledge that childhood is fleeting, and imaginative battles and heroic deeds will give way to a life of conference calls, political opinions, and concerns about the stock market.

Accompanying Weinersmith’s imaginative alliterations are Boulet’s wonderful illustrations, not only illustrating the tale but visually echoing the humor and joy of the words. While it’s a perfect book to read aloud to a child, adults will revel in it on their own, as they recall the imaginative fancies of their own childhoods.

While by no means a “faithful translation” (it would be no fun if it were), Bea Wolf does parallel the first third of the Old English epic, the fitts describing the hero’s battle with the monster Grendel. (Yes, the swimming contest and Unferth’s flyting are there too.) And while it does not attempt to parallel the entire poem, it does end with the appearance of Grindle’s mother, leaving the story open for children (and adults) to imagine what great battles and trials will come next.

The book is simply a joy, and I cannot recommend it more highly.

Weinersmith is the author of the internet cartoon Saturday Morning Breakfast Cereal (which I also highly recommend). Boulet is a French cartoonist and illustrator.

Discuss this post

Who is the bigger liar? George Santos or ChatGPT?

A color photograph of a man in glasses, wearing a blue suit jacket, sweater, and tie. An American flag is in the background.

Official portrait of George Santos

22 February 2023

I asked ChatGPT to write my biography. In doing so, it showed itself to be a worthy rival of George Santos.

Things I didn’t know about myself:

  • My birth certificate is wrong. Evidently, I was born on 28 November 1955 and not on a different day in 1963.

  • I was born in Chelmsford, Essex in the UK, and not in New Jersey, USA as my parents told me. (I don’t remember the event, myself.)

  • I attended the University of Sheffield, earning a bachelor’s degree in French and German. Although I do seem to vaguely recall attending Lafayette College in Pennsylvania where I did take a few semesters of German while majoring in Political Science. Don’t remember any French though.

  • I have a master’s degree in applied linguistics from the University of Edinburgh. Oddly, I always thought my master’s was in National Security Policy from George Washington University and that I have a PhD in medieval English from the University of Toronto. I do recall visiting Edinburgh, however.

  • Evidently, I worked as a lexicographer for both Chambers Harrap and Oxford University Press, and I was a senior editor for the second edition of the OED. Strangely, I always thought I was in the US Army stationed in Germany at the time the OED2 was published.

  • I was made an Officer of the Order of the British Empire (OBE) in 2016. I think would remember having met the queen, but maybe not.

Things I did know about myself:

  • I do work in the fields of linguistics and lexicography.

  • I’m the author of Word Myths: Debunking Linguistic Urban Legends.

  • I write the Wordorigins.org blog.

When I typed the request, I thought this would be a neat little experiment in the limits of what ChatGPT knows. I’m essentially a nobody, and I assumed—correctly as it turns out—that there would be little information about me in ChatGPT’s training data. I wanted to see what the AI did with just that kind of limited info. I suspected it would produce a couple of lines, mentioning the book and website, and that’s all. What I got surprised me.

First, it heaped on the praise. Evidently, I’m a “beloved figure in the world of linguistics and lexicography.” While I do like to think that’s the case, even I have to admit that the AI was laying it on thick.

But that’s not the worst of it. The AI spit out 334 words in 6 paragraphs of almost complete fiction. It got only two things right, the book and blog, just what I figured might be somewhere in its training data. What astounded me was the extent to which it fabricated “facts.” (An OBE?! Really?!)

I asked it to regenerate the biography listing its sources, and it cited three, giving URLs for each. The first was a non-existent Wikipedia article. The second was to a UK National Archives page on Brook Walton, an eighteenth-century English baronet and MP. (I had never heard of him, either.) And the third was to a non-existent page on the OED site. Not only did it create “facts,” it created fictional sources for them.

This was truly disturbing and an indicator of how the AI cannot be relied upon, at least not in its present state.

The fabrication of data is not simply a result of its training data being limited. The problem is in the very nature of its programming. When it has little data, it tries to generate plausible information rather than admitting that its knowledge is limited. It is easy to say that the AI will improve with more and better training data—I suspect a ChatGPT biography of Barack Obama, for instance, would be pretty darn accurate. While that’s undoubtedly true, it misses the point. The purpose of the AI is make connections between disparate data; it engages in induction. But an essential requirement of induction is recognizing the limits of one’s data and not to exceed them. ChatGPT is clearly terrible at this. More and better training data will reduce the frequency of this occurring, but it will always happen when its training data does not cover the subject at hand.

In this case, the fabrications are obvious (but perhaps not to a person who stumbles upon Wordorigins.org and wants to see my credentials, which might be the only reason someone, other than me, would want to ask an AI about me), but this is an insidious problem that undoubtedly occurs often but goes unnoticed in output for which the AI has good training data but fills in minor gaps with plausible-sounding fabrications.

Unless the AI can be taught to recognize the difference between induction and a lie, it is pretty much useless as a general tool. And those in Silicon Valley should think long and hard before releasing this or any similar AI into the wild.

Discuss this post

Image credit: U.S. House Office of Photography, 2022. Wikimedia Commons. Public domain image.

ADS Word of the Year for 2022

Word cloud of the nominees for ADS Word of the Year

The nominees for ADS Word of the Year

7 January 2023

Yesterday, the American Dialect Society voted on its Word of the Year (WOTY) for the past year. It selected the suffix -ussy, derived from pussy (as in bussy = boy pussy). The ADS uses a loose definition of word in its selection; any lexical item, as well as emojis and other signs, qualify.

The PDF of the ADS press release, listing all the categories, nominees, and vote totals, is here. There are many WOTY processes conducted by many different organizations, but the ADS selection is the original, having been conducted for thirty-three years.

The ADS consists of (mostly) professional linguists and lexicographers, but the selection of WOTY is not an academically rigorous process. Unlike other organizations that have some sort of objective criteria (e.g., Merriam-Webster bases its choice on searches in its online dictionary), the ADS process is informal. The evening before the vote, a small, self-selected group meets and comes up with nominations in the various categories. The categories are mostly consistent from year to year, but special categories can be created if there are a cluster of words on a particular topic. The next night several hundred conference attendees vote on the nominees, and nominations can be made from the floor. The process is raucous and fun, but the results can be skewed in all sorts of ways and should not be taken as serious and deliberative pronouncements.

I’ve participated in the nomination and voting in past years but sat out this year. (If I have another reason to attend the conference, or if it is online as in the past few years, I take part.) My own Wordorigins.org selections for Words of the Year are here.

What follows are my observations on the nominees and choices. Like the WOTY voting itself, my opinions are not serious linguistic conclusions. Pretty much all the WOTY contests, no matter who conducts them, are simply entertainment for the lexically inclined among us.

As for the overall WOTY choice, I’m unfamiliar with -ussy and the process of -ussification, but I don’t frequent TikTok, which evidently is where the term thrives. I would have gone with quiet quitting, which I predict will have more staying power and which is more representative of the mood of the year. Quiet quitting did win the category of Most Useful/Most Likely to Succeed. I concur with that choice, although another nominee, nepo baby, was also a strong contender in my book. -ussy also took the prize in the Most Creative category.

The ADS selection for Political WOTY was Dark Brandon, the supposed sinister, alter ego of Joe Biden, a play on the right-wing catchphrase Let’s Go Brandon. Again, I would have to disagree. Dobbs, a reference to the misogynistic US Supreme Court decision eviscerating women’s rights, received the second-most votes, and I predict it will long outlast Dark Brandon.

The choice for Digital WOTY was the suffix -dle, taken from the name of lexical games patterned after Wordle (e.g., Heardle, Absurdle, and Lewdle). That’s not a bad choice; games like this were quite the fad in 2022, and it is productive and creative. But there were several strong contenders in this category, such as chronically online, touch grass (and antidote for being chronically online), and crypto rug pull. Any of these would have been worthy choices, too. Chief twit was also nominated, and while it certainly held sway over much of the online world this past year, the less said about Elon Musk the better.

It's giving X was the choice for Informal WOTY. It’s another one that I’m unfamiliar with. It comes out of drag culture, which I am not plugged into. But it seems like a worthy choice, especially since the other nominees in this category were lackluster.

ADS’s Euphemism of the Year is special military operation, Putin’s name for the Russian invasion of Ukraine. I can’t disagree with this one. You don’t get more euphemistic than this.

The Snowclone/Phrasal Template of the Year went to not X, an ironic expression of mock horror or incredulity. This one is fine, and again I’m not familiar with it. (I’ve probably heard it but just didn’t register it.) I would have gone with X hits different, which garnered the second highest vote total, though.

Finally, Emoji of the Year went to the skull emoji 💀, used to express figurative death (e.g., from laughter or embarrassment). That’s a solid choice.

My biggest takeaway from this year’s choices is that I’m getting too old and too divorced from trends in popular culture. Most of the nominees were completely new to me. 💀

Discuss this post

2022: My Year in Astrophotography

31 December 2022

In case you didn’t know, I not only do words and historical linguistics, I’m also an amateur astrophotographer; that is, I take pictures of the night sky. I’ve been doing it, off and on, since 2008, but 2022 is the year that I finally got good at it. This is a compilation of the images I’ve taken during the past year.

I post my images to the Astrophotography section of the Wordorigins website and to Astrobin.com. If you want all the technical details about how I took the images, Astrobin is the place to find them, along with images taken by amateur astrophotographers around the world.

All of these images were taken from my driveway in Princeton, New Jersey, under Bortle 6 (bright suburban) skies.

I started off the year with an image of the:

26 January 2022

An emission nebula in the constellation Monoceros, 5,200 light-years away. This is a false-color image using narrowband filters to cut out light pollution. Emission nebulae consist of ionized gases that emit their own light. The filters cut out all light except that emitted by specific elements. This image uses the “Hubble palette,” assigning light from Hydrogen-ɑ gas to the green channel, from Sulfur-II to the red, and from Oxygen-III to the blue. The color scheme gets its name from its use by scientists using the Hubble space telescope.

This image is a total of 4.17 hours integration time (50 exposures of 5 minutes each), taken with a TeleVue NP127is refractor and a ZWO ASI2600 cooled, monochrome, CMOS camera mounted on my Takahashi EM-200 equatorial mount. I used a set of Baader 6.5-nanometter narrowband filters (H-ɑ, O-III, and S-II). My guide scope was a Sky-Watcher Evoguide 50DX with a ZWO ASI120MM Mini camera (which is the guide scope I’ve used for most of the images here). My capture software running the mount and camera is TheSkyX. Postprocessing used Deep Sky Stacker and Photoshop CC software.

Astrophotography is a challenging hobby. There are three main types of astrophotography: landscape or Milky Way, planetary, and deep-sky. All take practice, use different types of equipment, and require very different techniques. I haven’t done landscape work, which usually combines images of the Milky Way with compelling, earthly foreground objects. It uses a standard camera and tripod, or maybe a star-tracker mount for slightly longer exposures. I do a bit of planetary photography, which I describe below.

A telescope sitting in the driveway of a house

But most of what I do is deep-sky work, nebulae, galaxies, and star clusters. Deep-sky photography requires a mount that tracks and counteracts the rotation of the earth, a telescope, and a camera used for long (several minute) exposures that are integrated into a single, final image. Most people ask what telescope I use, but the most important piece of equipment is the mount. Unless you have a mount that accurately counteracts the rotation of the earth, you get blurred images. Also, a second, smaller telescope, camera, and computer are usually required for autoguiding. The second scope and camera focus on a single star and if it moves, send commands to the mount via a computer to correct for that movement. If done right, it results in an image with round, pinpoint stars with no elongation or trailing. And since the target objects are very dim, the images require considerable postprocessing work to reveal all their glory. (But without adding anything; if you see it in the image, it exists in the night sky.)

Here is the mount that I used for most of these 2022 images, a Takahashi EM-200. On it is mounted my TeleVue 127-mm refractor (660-mm focal length), dew heaters, computer, cooled camera, and filter wheel. The smaller telescope and camera mounted on top comprise the autoguider. I also have a 203-mm Ritchey-Chretien telescope (1,624mm focal length) and a 51-mm refractor (250-mm focal length). A long focal length produces magnified images of small objects, while a short focal length gives a wide-field view. I’ve had the Takahashi mount since 2008, and it has been a champ, but I’ve recently retired it in favor of a heavier-duty and more accurate Paramount MYT. More on that toward the end.

A cloud of brownish-yellow gases with a bright, white core against a starry sky

5–6 February 2022

The Orion Nebula (M42) is some 1,300 light years away. It is the middle “star” in Orion’s sword. This is one of the most popular deep-sky targets, but it is deceptively challenging. It’s easy to find—you can see it with the naked eye even in very light-polluted skies—but the high dynamic range (combination of faint gases on the outside of the very bright core) make it a tough one to image and process.

This is an exposure of 100 minutes of integration time (20×5-minute). Equipment is as above; except I started using PixInsight software for postprocessing instead of Deep Sky Stacker and Photoshop.

A spiral galaxy against a starry sky

12 & 29 April 2022

Spiral galaxy in Ursa Major, approximately 12 million light-years distant. This is a true-color image, using red, green, and blue filters with monochrome cameras.

This image has 4 hours, 25 minutes of integration time. (55×300sec). This time I used my 203mm TPO Ritchey-Chretien telescope to zoom in on the galaxy. I also used two different cameras for this one: ZWO ASI2600 & 1600 cameras. The 2600 is a superb camera, but a defect in quality control resulted in many of them suffering from oil leaking onto the sensor. My camera was out of commission for a while until I acquired the proper cleaning supplies. The problem kept recurring, and at the end of the year ZWO agreed to replace it. Hence, I reverted to my older ASI1600 for this and other images. Such is this hobby; seldom does everything work perfectly at the same time. That’s what keeps you on your toes.

A spiral galaxy drawing stars and gases away from a smaller galaxy

10 & 11 May 2022

Spiral galaxy in Canes Venatici, approximately 31 million light-years distant, and its companion NGC 5195, with which it is interacting. 

Integration time of 7.5 hours (90x300sec). Again, using my Ritchey-Chretien telescope.

Two nebulae, one shaped like North America and another that resembles a pelican

4, 5, & 9 June 2022

Emission nebulae in Cygnus. This is a wide-field image, taken with my William Optics Redcat 51 refractor telescope, focal length of 250mm for a wide-field view, and my ASI2600 camera. This is another narrowband image using the Hubble palette. But for this one (and all the ones going forward), I used a new set of narrowband filters, Antlia 3.5-nanometer H-ɑ, OIII, and SII filters.

Wispy tendrils of gases against a starry sky

14 June 2022.

A supernova remnant in Cygnus, some 2,400 light years distant.

3 hours, 45 minutes integration time. Again, using the Redcat 51, ASI2600, and narrowband filters.

EAGLE NEBULA (M16)

A bird-shaped nebula with pillars of gas at the center

24, 25, 30 June, 3, 10, 14 July 2022  

An open cluster with star-forming nebulae in the constellation of Serpens, some 5,700 light-years distant

This was my longest integration time to date, a total of 22 hours, 52 minutes, 30 seconds integration time, using 450-second exposures of H-alpha, Oxygen III, and Sulfur II. This time I went back to my TeleVue NP127is refractor, which gives me a medium field of view between that of the Ritchey-Chretien and the Redcat 51.   

Close-up of the pillars of gas in the above image

Here is a detail of the above image, showing the “Pillars of Creation,” a star-forming region, in close-up.

Clouds of brown-colored gases around a bright star

19–22 July 2022

Emission nebula surrounding the star Sadr (Gamma Cygni) in the constellation Cygnus. Another mega-integration time: 25 hours, 37 minutes, 30 seconds; 450-second exposures of H-alpha, Oxygen III, and Sulfur II. Equipment as above.

Starless image with a single pillar of gas that resembles an elephant's trunk at the center

3 August 2022

A starless take on the nebula in the constellation Cepheus, about 2,400 light years distant. I’d been trying to get the backfocus right on my telescope. Backfocus is the distance between the last optical element in the imaging train and the camera’s sensor, and if it’s not right, the stars at the edges can appear misshapen. So, this is an image with the stars removed. It has the advantage of making the nebulosity more prominent. It’s common to remove the stars during postprocessing in order to work on the nebulosity, but usually you add the stars back.

Total of 6 hours, 37 minutes, 30 seconds integration time; 450-second exposures of H-alpha, O-III, and S-II. Equipment as above.  

A gas-giant planet with layers of gas running across it

8 September 2022

My first serious attempt to image a planet. Planetary imaging is a very different beast, using a technique called lucky imaging. Atmospheric distortions wreak havoc on planetary images; deep-sky targets are much more forgiving in this respect. To combat this, one takes a video—thousands of frames—and combines only those “lucky” few that happen to be clear into the final image. I’m still learning.

This one used my Ritchey-Chretien scope and a ZWO ASI290MM camera, with infrared, red, green, and blue filters. I used the infrared frames as a luminance channel. Postprocessing was with Autostakkert and RegiStax software.

A gas-giant planet with rings around it

8 September 2022

Saturn on the same night as Jupiter with the same equipment, only I also used WinJupos software in postprocessing. This one came out much better.

A large, spiral galaxy against a starry sky

15 September 2022

Widefield shot of the nearest major spiral galaxy, 2.5 million light years distant. Andromeda is massive, taking up about 3 degrees in the night sky—that’s 6 full moons across, but so faint that you only see the bright core with the naked eye, if light pollution permits even that.

This one was taken with my portable rig: my Redcat 51 refractor and Canon D6 Mark II (modified) DSLR camera, mounted on a Sky-Watcher StarAdventurer GTi star tracker. The camera is modified to extend the spectrum it captures deeper into the red so it can capture Hydrogen-ɑ emissions. I also used an Optolong L-Pro light-pollution filter.

Total of 1 hour, 9 minutes integration time (23×180-second exposures)    

A cloud of red and teal gases that resemble a human eye against a starry sky

23 September 2022

Also called the Eye of Sauron, this is a planetary nebula some 650 light-years distant in the constellation Aquarius. Planetary nebula is a misnomer, dating to the eighteenth century when astronomers thought these objects resembled planets. They are formed by dying stars sloughing off gases.

For this one I used a cooled, color, astronomy camera, a QHY183C, my Optolong L-Pro light-pollution filter, and my TeleVue NP127is refractor. Integration time was 3 hours (36×300 seconds).

HEART & SOUL NEBULAE (IC 1805 & IC 1848)

Two red nebula, one resembling a human heart, against a starry sky

23 September 2022

Emission nebulae in Cassiopeia, some 7,500 light-years distant.

I ran two different imaging rigs this night. This second one was taken with my portable rig, using an Optolong L-Pro filter. 5 hours, 56 minutes, and 16 seconds of integration time (167×128 seconds).

A spiral galaxy against a starry sky

27 September 2022   

This is the third spiral galaxy in our local group, the others being Andromeda and our own Milky Way. At 2.73 million light-years, it's the most distant object that can be seen with the naked eye.

Total of 8 hours, 20 minutes integration time (100×300 seconds) with LRGB filters, using my TeleVue NP127is refractor and ASI2600 camera.    

Two nebulae, one resembling a lobster's claw and the other containing what looks like a soap bubble

8–11 October 2022

Emission nebulae in Cassiopeia, some 7,100 to 11,000 light-years distant

Total of 26 hours, 45 minutes integration time, 214×450-second exposures of Hydrogen-α, Oxygen-III, and Sulfur-II, using my TeleVue NP127is refractor and ASI2600 camera.

Here is a detail of the above image showing the “bubble.”

LUNAR ECLIPSE

A composite of images of the full moon, running from gray on the left, gradually disappearing in to blackness, until the final image of a red moon

8 November 2022

Composite of images from the first half of the eclipse. I didn't get the second half because of sunrise. For this one, I just used a tripod, my Canon 6D Mark II with my Redcat 51 telescope as the lens, and a 4× Barlow lens for magnification. Various exposure times at ISO 1600.

Gray and orange nebulae with a cloud of black dust resembling a horse's head at the center

4 December 2022

A telescope sitting in a driveway

Dark and emission nebulae in Orion; the bright star is Alnitak, the leftmost star in Orion’s belt. Total of 6 hours, 30 minutes integration time, 39×10-minute exposures of Hydrogen-α, Oxygen III, and Sulfur II. 

This image was also processed using Russell Croman’s BlurXterminator tool, which is an AI program that conducts deconvolution, sharpening stars and nebulae. Deconvolution is a laborious and painstaking process that is difficult to get right. This tool makes it simple. I expect it will improve my images no end going forward.

This image is also the first light using my new mount, a Paramount MYT equatorial mount. Here is my Paramount with my TeleVue refractor mounted on it. The red camera at the end is the ASI2600, which is attached to a filter wheel that holds the various filters I use. The red box on top is the SkyX Fusion computer that runs the rig. Since this picture was taken, I’ve added a Moonlite Nitecrawler electronic focuser, which keeps the telescope in focus as conditions change throughout the night.

Orange nebula that is said to resemble a map of California

9 December 2022 

Emission nebula in Perseus, some 1,000 light-years distant. Some people think this nebula resembles a map of California, hence the name; I don’t see it, but to each their own. A total of 9 hours integration time, 27×20-min exposures of Hydrogen-α, Oxygen III, and Sulfur II. This one used my Redcat 51 on the Paramount MYT, and my ASI1600 camera (my 2600 having been sent back for replacement due to the dreaded oil leak problem).  

Amazingly, the Paramount managed 20-minute exposures with no star trailing and did this without the help of a second guide scope. The Redcat has a wide field of view, which is more forgiving, but even so, this is amazing performance in a mount. I can’t wait to see what this mount allows me to do in 2023.

An orange nebula with a teal center

12–13 December 2022

My final image of the year: the emission nebula in Cassiopeia which I had previously imaged above alongside its companion, the Heart Nebula. A total of 17 hours integration time, 51×20-min exposures of Hydrogen-α, Oxygen III, and Sulfur II, using the Redcat 51 and ASI1600 camera. Again, no guide scope or camera.

I’m looking forward to what 2023 will bring, and I’m wishing you all the best for the new year. 

Discuss this post